We tested a computer system for routine arrhythmia monitoring in 181 postoperative patients in order to evaluate its practical value. There was no significant difference in the incidence of arrhythmias between patients clinically classified in poor condition and those classified in good condition. Conventional vital signs appeared to have more value than a PVC count or an index of arrhythmias. We found no evidence that continuous electrocardiogram monitoring might provide an early warning signal in these patients monitored.